The Effects of Non-Binding Promises on Sequential Cooperation

\(15^{th}\) BEEN Conference - LUISS, 23 January 2026

Sandro Casal

DEM-CEEL, University of Trento

Luigi Mittone

DEM, University of Trento

Matteo Ploner

DEM-CEEL, University of Trento

Motivation

Chains that depend on trust

  • Many interactions are sequential handoffs where early actors invest only if later actors promise to continue.
    • Sustainable supply chains, multi-stage projects, sequential teamwork…
  • Non-binding promises are common in such contexts (e.g., verbal commitments, emails).
  • Question: can non-binding promises sustain cooperation in sequential chains?

Research question

Do non-binding promises increase cooperation in a five-player sequential chain of cooperation?

Mechanisms:

Promise incidence 💬✔️

How likely is a player to send a pass promise?

Conditional responsiveness ↪️🤝

How likely is a player to cooperate after receiving a promise?

Key idea
In sequential chains, policy leverage may lie in eliciting explicit commitments, not in richer communication.

Previous contributions 🐛

Centipede cooperation

Promises and cooperation

Design

Five-player centipede game

  • Players \((P_1,\dots,P_5)\) move sequentially.
  • At node (i): choose Take (stop) or Pass (continue).
    • Take = immediate payoff to self; others get 0.
    • Pass = cooperation to reach final node.
  • If all pass: equal split of total potential take-payoffs.

Treatments (between-subjects)

Treatment Promise (available?) Mandatory promise (Take/Pass) Fee for sending message
Baseline (B) ✗ ✗ ✗
Cheap Talk (C) ✓ ✓ ✗
Voluntary (V) ✓ ✗ ✗
Fee (F) ✓ ✗ ✓

Hypotheses

  • H1 (Promises): take-up of pass promises differs by regime.
    • H1a: Pass promises same in Voluntary and Cheap Talk.
    • H1b: Pass promises lower in Fee than Voluntary.
  • Motivation: promises are disciplined by lying psychological costs; when talking is costly, fewer promises are sent.

Hypotheses

  • H2 (Cooperation): any promise regime increases Pass vs Baseline.
  • Motivation: allowing promises should raise commitment and reciprocity along the chain vs baseline with no communication.

Hypotheses

  • H3 (Credibility/response):
    • H3a: Pass promise triggers more passing in Voluntary than Cheap Talk (silence makes it “less coerced”).
    • H3b: Pass promise triggers more passing in Fee than Voluntary (fee increases credibility).
  • Motivation: the more voluntary/costly the message, the more credible a Pass promise should be, so recipients should respond more cooperatively.

Participants and procedure

  • Preregistered on OSF (DOI: 10.17605/OSF.IO/NYT2U): design, hypotheses, sample, analysis
  • Recruited on Prolific (US adults); English-fluent, ≥ secondary education, 90–100% approval
  • N = 3003 after exclusions (B=753; C=750; V=750; F=750)
  • Payoffs: £1 show-up + bonus (0.05 £/point); median ~3 min; avg £20.27/hour
  • Procedure: oTree online; session-level randomization; strategy method + (C/V/F) promises + unincentivized beliefs; ex post matching into 5-player chains

Results

R1 — Promise incidence (H1)

  • Pass promises: \(C > V > F\)
  • No-message substantial in Voluntary and especially in Fee

R2 — Aggregate cooperation (H2)

  • Highest cooperation in Cheap Talk
  • Voluntary and Fee: close to Baseline

R3 — Conditional responsiveness (H3)

  • Pass promise received → Pass choice increase
  • Conditional response is similar across treatments (conditional responsiveness stable)

Interpretation

Incidence and responsiveness

Promise incidence (how often promises are sent)

  • Institutions change how often the promise channel is activated.
  • Mandatory, costless messaging ⇒ many explicit commitments.

Conditional responsiveness (effect of receiving a promise)

  • A pass promise is behaviorally meaningful in all regimes.
  • Responsiveness is not markedly different across regimes in the data.

Takeaway
Aggregate cooperation is driven primarily by the prevalence of explicit commitments, not by large credibility shifts conditional on a promise.

Implications for sequential processes

  • In handoffs / workflows, allowing communication is not enough.
  • Active elicitation (required acknowledgments, commitment prompts) can outperform optional tools.
    • Promises are largely honored and reciprocated once issued.
    • Small frictions (fees, hassle costs) can shut down the commitment channel.

References

Abeler, Johannes, Daniele Nosenzo, and Collin Raymond. 2019. “Preferences for Truth-Telling.” Econometrica 87 (4): 1115–53.
Balliet, Daniel. 2010. “Communication and Cooperation in Social Dilemmas: A Meta-Analytic Review.” Journal of Conflict Resolution 54 (1): 39–57. https://doi.org/10.1177/0022002709352443.
Battigalli, Pierpaolo, and Martin Dufwenberg. 2007. “Guilt in Games.” American Economic Review 97 (2): 170–76.
Charness, Gary, and Martin Dufwenberg. 2006. “Promises and Partnership.” Econometrica 74 (6): 1579–1601. https://doi.org/10.1111/j.1468-0262.2006.00719.x.
Fey, Mark, Richard D McKelvey, and Thomas R Palfrey. 1996. “An Experimental Study of Constant-Sum Centipede Games.” International Journal of Game Theory 25 (3): 269–87.
Krockow, Eva M, Andrew M Colman, and Briony D Pulford. 2016. “Cooperation in Repeated Interactions: A Systematic Review of Centipede Game Experiments, 1992–2016.” European Review of Social Psychology 27 (1): 231–82.
McKelvey, Richard D, and Thomas R Palfrey. 1992. “An Experimental Study of the Centipede Game.” Econometrica: Journal of the Econometric Society, 803–36.
Rapoport, Amnon, William E Stein, James E Parco, and Thomas E Nicholas. 2003. “Equilibrium Play and Adaptive Learning in a Three-Person Centipede Game.” Games and Economic Behavior 43 (2): 239–65.
Vanberg, Christoph. 2008. “Why Do People Keep Their Promises? An Experimental Test of Two Explanations 1.” Econometrica 76 (6): 1467–80.